Analysis of Microsoft’s FY24 Q2: You Can’t Spell Activision Without AI (Premium)

Yesterday, Microsoft reported a net income of $21.9 billion on revenues of $62 billion for the second quarter of its fiscal year 2024. Here’s a more detailed look at this quarter that draws on the additional information Microsoft provided during its post-earnings conference call, with the usual emphasis on the consumer businesses that we care about the most here on Thurrott.com.

The raw numbers

• Overall earnings: Net income of $21.9 billion (up 33 percent year-over-year) on revenues of $62 billion (up 18 percent YOY)
• Intelligent Cloud: $25.9 billion in revenue, up 20 percent YOY
• Productivity and Business Processes: $19.2 billion in revenues, up 13 percent YOY
• Personal Computing: $16.9 billion in revenues, up 19 percent YOY

The impact of Activision Blizzard

More Personal Computing houses Windows, Xbox, and Surface, and while it has long been the smallest of Microsoft’s business units by revenue, it has suffered more dramatically in the post-COVID era thanks to falling PC and Xbox console sales. So it was not surprising that More Personal Computing brought up the rear yet again in terms of revenues. But the gap between it and Productivity and Business Processes closed significantly, thanks entirely to Microsoft’s acquisition of video game giant Activision Blizzard.

This is, I think, the biggest news of the quarter.

That said, it wasn’t entirely unexpected: Last July, I ran the numbers on how Activision Blizzard would impact More Personal Computing specifically and Microsoft more broadly, and my conclusion from that time was that Activision would improve More Personal Computing’s standing against Microsoft’s two other business units, but not even to make it the second-biggest Microsoft business unit. As for Microsoft generally, I expected Activision Blizzard to contribute to 4 percent-ish in additional revenues in any given quarter (above whatever other factors contribute to those numbers).

And that’s exactly what happened.

Had Microsoft not completed its acquisition of Activision Blizzard, More Personal Computing revenues would have grown just 4 percent YOY to $14.25 billion, according to the numbers Microsoft provided. And Microsoft’s total revenues would have been just $59.3 billion, a difference of just … wait for it … 4.3 percent. Look at me doing math!

But Activision’s contribution to More Personal Computing is enormous, with over $2 billion in additional revenues overall. Consider that Xbox content and services revenues grew by 3, 5, and 13 percent in the previous three sequential quarters, but then grew by 61 percent in this Activision Blizzard-infused quarter. Had this acquisition not occurred, Xbox content and services revenues would have grown by less than 13 percent, and that was a holiday quarter.

There are expenses related to the acquisition, of course, and, no, I’m not talking about the $69 billion that Microsoft paid for this company, as that was accounted for long ago. Activision Blizzard employed nearly 10,000 people across its many studios at the time of the acquisition, and when Microsoft merged that business into its Microsoft Gaming division, the employee count there ballooned to 22,000, with many redundancies. This explains the reorgs we’ve seen since the acquisition. But it also explains the recent layoffs: Before the layoffs, Microsoft Gaming’s operating expenses were 38 percent higher than in the year-ago quarter. I doubt we’ll see how this changes over the coming year—Microsoft likes to cherry-pick which numbers it reports—but Microsoft Gaming will always be bigger than it used to be, and that comes with additional costs.

There are other, smaller costs associated with this acquisition, too. When Activision Blizzard was a standalone company, it was a Microsoft partner that contributed to the third-party revenue that it collects from game publishers who sell content through Xbox. But now that it’s a first-party publisher, Microsoft realizes all of Activision Blizzard’s revenues, of course. Less obviously, it loses out on the third-party revenues it used to collect.

We don’t know what those numbers look like. But Microsoft provided an interesting look at the net impact of Activision Blizzard on the company, and when you factor in the integration costs, transaction-related costs, and other operating expenses related to Activision Blizzard, you get a net loss of $440 million in the quarter. But that’s a one-time thing, and the goal of the reorgs and layoffs is to iron out the accounting and let the integrated Activision Blizzard sink or swim.

There were other interesting numbers.

Thanks, again, to Activision Blizzard, Microsoft now has over 200 million monthly active users on Xbox, PC, and mobile, a record. By comparison, Microsoft reported that it had 150 million monthly active users in July 2023.

Hours spent streaming games via Cloud Gaming was up by 44 percent YOY, a meaningless number given that it’s a small market to begin with, and we have no hard user numbers.

Microsoft noted that the “gaming console market was a bit smaller” in the quarter. I’m not sure what that means, but Xbox console revenue grew 3 percent in the quarter.

Moving on.

Windows rises with the PC market

Before getting into the meat of the quarter, Microsoft discussed a few superfluous factoids about Windows.

First, it claimed that “Copilot in Windows is already available on more than 75 million Windows 10 and Windows 11 PCs.” I’m surprised that number isn’t higher, since its arrival on Windows 10 puts it in front of an audience of about 1 billion people.

Then it noted that Windows 11 commercial deployments were up 100 percent (2X) year-over-year thanks to big companies like HPE finally rolling out the platform. But this doesn’t impact revenues: Microsoft gets paid the same each month, no matter which supported Windows version its commercial customers use.

So here’s what really matters.

In keeping with the numbers we saw earlier from Intel and AMD, the PC market had a bit of a rebound in the final quarter of 2023, though that’s a low bar given how horrific the past two years have been for the industry. And the revenue figures that Microsoft provided bear that out: Windows revenues from PC makers jumped by 11 percent in the quarter, Microsoft says, after a 4 percent gain in the previous quarter and 12 and 28 percent declines in the two before that. The upward bend of that curve is slight but obvious, and it’s heading in the right direction.

Sadly, Microsoft expects PC revenues to be flat in the current quarter “as PC market unit volumes continue at pre-pandemic levels.”

Windows revenues from commercial (i.e. business) products and cloud services was also up, in this case by 9 percent, compared to 8, 2, and 14 percent growth in the previous three sequential quarter. These numbers are impossible to predict because they’re based on unknowable business upgrade cycles. But it’s worth pointing out, I think, that this part of the business suffered a 3 percent decline in the year-ago quarter. So it was a good year overall.

Just not for Surface.

Surface rises not with the PC market

Surface continues to be a drag on the company. Microsoft’s Devices business, which is basically just Surface and HoloLens or, objectively just Surface, saw a revenue decline of 9 percent in the quarter, and I can’t recall the last time it saw growth. The previous four sequential quarters suffered revenue declines of 22, 20, 30, and 39 percent, respectively, and I can’t bear to look back any further than that. But the ongoing woes at Surface explain why Microsoft simplified the product line last year to focus on “higher margin premium products” and, most likely, why Panos Panay left the company.

I don’t see an end to this bloodshed, and I think we’re firmly in the “it’s not if, it’s when” part of the discussion around Microsoft just giving up. Activision Blizzard has raised the cost of operating More Personal Computing enough to justify the massive savings that Microsoft will enjoy by killing Surface. We’ll see.

Speaking of which.

The cost of AI

There’s an interesting note attached to Microsoft’s search and news advertising business, which is a part of More Personal Computing that I’ve largely ignored over the years because there is nothing interesting about Bing, MSN, Microsoft Start, and the software giant’s advertising business. Except, of course, that there is now: Last year’s explosive AI initiative was notable for many reasons, but one of the many big stories there is that Microsoft in February 2023 decided to base its initial go-to-market strategy on Bing, one of its least successful brands and products. A second and very much related big story is that the software giant quickly reversed course on this strategy after it became clear that stuffing Bing full of AI did absolutely nothing to improve its usage, revenues, or profits. Today, Microsoft markets these efforts under the Copilot brand instead, heavily de-emphasizing Bing both as a brand and a destination: You can access Copilot all over the place now.

In July 2023, I examined the impact of AI and wondered how Microsoft would pay for this extremely costly technology. And in October 2023, we got an answer: Microsoft was paying somewhere between $10 billion and $15 billion per quarter to build out its AI infrastructure, and those costs were only going to go up moving forward as the AI arms race continued. The good news was that Microsoft had the cash, given its enormous products. But the bad news was that this kind of expense is unsustainable. At some point, Microsoft’s shareholders and Wall Street will expect the company to collect money from AI, not just spend it.

This past quarter, these two realities collided, with Microsoft reporting that the cost of its search and news advertising business grew by 8 percent, driven by the combination of “higher search volume” and the “continued unfavorable impact from a third-party partnership.” This very clearly references OpenAI. And it speaks to why Bing is so broken: It’s not enough to have higher traffic when the cost of that traffic is so expensive.

Of course, Bing’s role in Microsoft’s broader AI push is modest. Most of the heavy lifting is happening in Azure, which is part of the Intelligent Computing business unit. And Microsoft’s disclosures about this Azure and AI are interesting.

Say what you will about Satya Nadella, but I find the man cold and robotic, and the way he speaks almost physically hurts my ears. So I was surprised to see that he spoke quite eloquently about AI, training, inferencing, reinforcement learning from human feedback, and related AI topics during the call last night. As did Amy Hood, if at a lower technical level. But the two paint a pretty AI cost-to-benefit ratio for Azure and Microsoft 365, specifically. And so it is not surprising that Microsoft is focusing its investments and efforts there. And not in Bing.

I just wrote about how Copilot Pro and Copilot for Microsoft 365 are basically new higher-end SKUs for Microsoft 365 (consumer and commercial), and Microsoft confirmed this when one caller asked about the maturation of the Microsoft 365 commercial market, where growth in the number of seats inevitably slows over time. In recent years, Microsoft 365 seat growth has come from “small medium-sized businesses and really, frontline workers scenarios,” but that growth is, of course, slowing. But now, most future growth will come from the paid Copilot subscriptions, but build on top of existing subscriptions (seats).

“[AI] has evolved in terms of its economic benefits or productivity benefits,” Nadella said in answering a question about the value of GitHub Copilot specifically. “It’s like if you take away spellcheck from Word, I’ll be unemployable. And similarly, it’ll be like I think GitHub Copilot becomes core to anybody who is doing software development.” If the Copilot subscriptions for Microsoft 365 ever get that good, the same rules will apply. That is, of course, the goal. Making paying more not just a no-brainer, but a requirement.

“I do see this as a new vector for us in what I’ll call the next phase of knowledge work and frontline work, even, and their productivity, and how we participate,” he added.

Welcome to our AI future.